2/11/20 15-292 History of Computing IBM Domination in the 1960s and 1970s IBM ● 1962 – first year that computers revenue exceeded that of punched-card machines ● By end of ’60s, no more punched-card machines ● ’60s – growth of 15-20% per year ● 1960 - $1.8 biLLion in saLes, 104,000 empLoyees ● 1970 - $7.2 biLLion in saLes, 259,000 empLoyees ● sustained 70% share of computers market throughout the decade ● The next great success: IBM’s 1401 1 2/11/20 IBM 1401 IBM 1401 ● A stored-program transistor-Logic computer system ● foLLow-up to 650 ● aimed to be cheaper, faster, & more reLiabLe ● transistors for vacuum tubes, core memory for magnetic drum ● high-speed printer (600 Lines/minute) ▪ heLped Lure customers using punched-card machines ● $2500 per month minimaLLy configured in 1960 ● It was the first computer to depLoy 10,000 units ● IBM concentrated on computer systems rather than just individuaL machines & architectures ● customer driven 2 2/11/20 Other 1401 Specs ● The 1401 was a decimaL (not binary) computer intended primariLy for business appLications ● OriginaLLy programmed onLy in Autocoder (assembLer) ● proved difficuLt for many peopLe ● Soon used one of the earLiest high-LeveL business- oriented programming Languages, RPG ● Report Program Generator ● increased its usabiLity and popuLarity IBM 1401 3 2/11/20 IBM 1401 IBM 1401 Reconstructed IBM 1401 reconstruction at the Computer History Museum 4 2/11/20 IBM Competition ● IBM and the Seven Dwarves – late 1950’s ● Sperry Rand (UNIVAC division) ● Control Data Corp. (CDC) ● RCA ● HoneyweLL ● GE ● Burroughs ● NCR ● IBM enters into a consent decree with the U.S. Government in 1956 agreeing to sell as welL as lease its computers ● Leads to many Leasing companies ● Provide parts to other companies to maintain IBM machines ● http://www.cptech.org/at/ibm/ibm1956cd.htmL ● Consent decree remains in effect untiL 1996 IBM Competition ● Companies Like HoneyweLL, GE, & RCA started to produce IBM-compatibLe machines ● to target IBM customers ● HoneyweLL’s 200 was an improvement over IBM 1401 ● 400 orders in first week (more for HoneyweLL than in previous 8 years) ● AdditionaL IBM competition: itseLf ● too many different machines (7 Lines) not fuLLy compatibLe ● a particuLar probLem with software ● not one of IBMs modeLs couLd run the software of another 5 2/11/20 SPREAD Task Group ● SPREAD: Systems, Programming, Review, Engineering And DeveLopment ● EstabLished by Vincent Learson in October 1961 ● Consisted of IBM’s 13 most senior engineering, software and marketing managers. ● “Banished” to Sheraton New EngLander in Cos Cob, Connecticut to come up with a new product Line of compatible computers ● Proposed a range of compatibLe computers that wouLd repLace alL of IBM’s existing computers – System/360 ▪ enormous, secret undertaking ● Software deveLopment estimate: $125 miLLion?!?!?!? ● Project nicknamed “You bet your company” by IBM engineers ● ResuLting direct research costs: $500 miLLion ● ResuLting deveLopment costs: $5 biLLion ▪ Second in ’60s onLy to ApoLLo project IBM System/360 ● “the computer that IBM made that made IBM” ● CaLLed 360 because of its “betokening aLL points of the compass” ● Suggesting universaL appLicabiLity of the machines ● An entire Line of computers ● smaLL to Large ● Low to high performance ● aLL (with but one exception) running the same command set ● Announced with much drama on ApriL 7, 1964 ● Watson Jr, “the most important product announcement in computer history” ● An immediate success, IBM couLd not fuLfiLL aLL the orders it got ● Some modeLs (e.g., the 360/30) even offered the option of microcode emulation of the customer's previous computer ● oLd programs couLd stiLL be run on the new machine 6 2/11/20 IBM System/360 IBM System/360 ● The most expensive CPU project in history. ● Fortune magazine: "$5 BiLLion gambLe" ● The System/360 introduced a number of industry standards to the marketpLace, such as: ● the 8-bit byte (against financiaL pressure during deveLopment to reduce the byte to 4 or 6 bits) ● byte-addressabLe memory (as opposed to word-addressabLe memory) ● 32-bit words ● segmented and paged memory ● commerciaL use of microcoded CPUs ● couLd be configured for networked appLications 7 2/11/20 IBM System/360 ● The S/360 famiLy initiaLLy consisted of six computers and forty common peripheraLs ● There were thirteen modeLs in aLL. ● The cheapest modeL was the 360/20 ● 24K of memory ● half the registers of other models ● the instruction set was not binary-compatibLe with the rest of the range ● The most significant modeL was the 360/67 ● first shipped in August 1966 ● the first to offer virtuaL machine computing to its users through its CP-67 operating system IBM System/360 IBM 360 Model 75 - 1965 (IBM Archives) IBM 360 Model 25 - 1968 (IBM Archives) 8 2/11/20 IBM System/360 IBM 360 ModeL 65 consoLe - 1965 (IBM Archives) IBM 360 Model 91 - 1968 (IBM Archives) IBM System/360 ● Some wouLd say the impLementation was “pedestrian” ● Ex: did not support time-sharing ● where muLtipLe parties/programs may share use of a machine ● IBM Audit said “IBM engineering was mediocre” ● Some wouLd say the marketing was far stronger than the implementation ▪ RCA wouLd immediateLy make 360-compatibLe clones 9 2/11/20 OS/360 & Fred Brooks ● A batch processing operating system deveLoped by IBM for the System/360 ● Versions: ● PCP – Primary ControL Program ● MFT – MuLtiprogramming w/ Fixed Number of Tasks ● MVT – MuLtiprogramming w/ VariabLe Number of Tasks ● DeLayed for over a year due to organizationaL disarray and inexperience in deveLoping large-scaLe software systems ● Frederick P. Brooks pubLishes The MythicaL Man-Month in 1975 describing the second-system effect Brooks Interview 10 2/11/20 Brooks’ Law Programming work performed increases with direct proportion to the number of programmers (N), but the complexity of a project increases by the square of the number of programmers (N2). Therefore, it should follow that thousands of programmers working on a single project should become mired in a nightmare of human communication and version control. Steve Jobs references The Mythical Man-Month Herbert Grosch ● In 1945, he was drafted into the new IBM Watson Lab at CoLumbia by Los ALamos to provide backup for bomb caLcuLations. ● Grosch’s Law (1965): Computer performance increases as the square of the cost. ● You have a computer that costs $100,000 ● Another computer that costs $500,000 wiLL be 25X as powerful. ● It is cheaper to buy one $500K computer for 25 peopLe than 25 $100K computers. ● His Law didn’t appLy in the 1970s as the cost of computer power shrank by a factor of 100 due to integrated circuits. 11 2/11/20 Herbert Grosch and a future U.S. President Ronald Reagan and Herb Grosch at an IBM 701 in 1954. Ronald Reagan becomes spokesperson for General Electric and their new computer ERMA: Electronic Recording Machine – Accounting Meant to speed up banking transactions Pressure on IBM (late 1960s) 12 2/11/20 IBM System/370 ● A Line of IBM mainframes to be the successor to the System/360 famiLy (announced 1970) ● Cheaper & better technoLogy than 360 ● used true integrated circuits (ICs) ● semiconductor RAM rather than core memory ● enhanced address space ● virtuaL memory ● DeveLoped first at Manchester University ● “As aLways, IBM’s pubLicity machine was stronger than its technoLogy.” – CampbeLL-KeLLy, Aspray IBM System/370 IBM System 370 ModeL 135 consoLe (IBM Archives) IBM System 370 ModeL 125 consoLe (IBM Archives) 13 2/11/20 IBM System/370 IBM Future Series (FS) ● Launched in the 1970s to make another major Leap to create a pLatform that wouLd have reduced software costs. ● PLanned for Late ’70s reLease ● In 1975, IBM stopped the project after many deLays. ● Reasons for faiLure: ● Vague objectives? ● Objectives too far ahead of avaiLabLe technoLogies? ● Poor management after Watson Jr. retired (1971)? ● Specter of existing software investment ● $100 miLLion for nothing, “the most expensive deveLopment-effort faiLure in IBM’s history” ● The 370/ESA was eventuaLLy rebranded as the System/390, and Later stiLL as the zSeries. 14 2/11/20 IBM System/390 & zSeries IBM System 390 (IBM Archives) IBM zSeries z990 (IBM Archives) The Decline of the IBM Empire ● Anti-trust suit ● More of a broadening of the market than a coLLapse of IBM ● StiLL one of the most profitabLe companies in the worLd ● Though it Lost $5 biLLion in 1992, more than any U.S. company had ever Lost in a singLe year ● StiLL the industry Leader in mainframe computers ● Emergence of mini & then micro computers ● Low-cost ICs aLLowed new companies to enter what was once an exclusive club ● As software advanced, companies needed Less of IBM’s service, which was their greatest asset ● Percent of gLobaL market: 1976: 50% 1985: 25% 15.